Jarnail Singh and B.K. Pavitra explained: the data test for promotion quotas

On 9 February 2017 the Supreme Court struck down two sections of a Karnataka law on promotions. Not because the officers promoted under it were unfit, and not because reservation in promotion is impermissible. Because of what had not been done before the law was passed:

In B K Pavitra I, this Court struck down Sections 3 and 4 of the Reservation Act 2002 as ultra vires Articles 14 and 16.

Ultra vires means beyond the power the Constitution gives the State, so the sections were not narrowed. They went. And the ground was an absence: the exercise of determining inadequacy of representation, backwardness and the impact on efficiency had never preceded the enactment.¹

Karnataka then did the exercise. A committee was constituted six weeks later and reported six weeks after that. On the strength of it the State won in 2019. In 2022 the Court held that the way it had counted was the wrong way.

Three boxes, and who has to fill them

The requirement comes from M. Nagaraj, which is the subject of the previous part. Article 16(4A) is an enabling provision, so nothing compels a State to reserve in promotion at all. The obligation only starts once it decides to:

The state is not bound to make reservations for the SCs and STs in promotions. But, if it seeks to do so, it must collect quantifiable data on three facets:

Those three are inadequacy of representation, backwardness, and overall efficiency, and the standard attached to them is not a low one.¹ Quantifiable means counted: numbers gathered and produced, rather than a belief that a group needs help.

Whether States could meet it is answerable from the record, and the answer is in a document written to complain about it. Parliament tried to remove the requirement at source: the Constitution (One Hundred Seventeenth Amendment) Bill, 2012 would have deemed Scheduled Castes and Scheduled Tribes backward, so a State would not have to prove it again. The Statement of Objects and Reasons attached to that Bill set out what had been happening in the courts:

Relying on the judgment of the Supreme Court in M. Nagaraj case, the High Court of Rajasthan and the High Court of Allahabad have struck down the provisions for reservation in promotion in the services of the State of Rajasthan and the State of Uttar

The judgment records that those decisions did not get quietly reversed on appeal.² That is the Bill's own account of why it was thought necessary, and the Bill did not pass.

This Bill was passed by the Rajya Sabha on 17.12.2012 but failed to get sufficient number of votes in the Lok Sabha and, therefore, could not become an Act.

The exit stayed shut.

One box removed, and another opened

On 26 September 2018 a Constitution Bench heard a request to send Nagaraj to seven judges. The Attorney General argued it needed revisiting on two points: the data requirement on backwardness, and the application of the creamy layer to Scheduled Castes and Scheduled Tribes.²

The Bench refused the reference and then changed Nagaraj anyway:

Thus, we conclude that the judgment in Nagaraj (supra) does not need to be referred to a seven–Judge Bench. However, the conclusion in Nagaraj (supra) that the State has to collect quantifiable data showing bac

The reasoning is short. Backwardness is presumed for these groups, so requiring a State to prove it again runs against the nine judges of Indra Sawhney.¹

That is one box gone. But the same judgment brought in something the Government had argued against, and it reopened a line Indra Sawhney had drawn. That judgment's creamy-layer discussion had carried an express limit: ¹ In 2018 the Court held that constitutional courts may exclude the creamy layer when applying the principle of reservation.² The bracket that had kept Scheduled Castes and Tribes outside the creamy layer in 1992 no longer did.

What Karnataka counted

The committee was constituted on 22 March 2017,¹ and reported on 5 May 2017.¹ Forty-four days for the exercise whose absence had brought down a statute.

Its numbers are worth reading with their denominator attached:

The total number of sanctioned posts as per the data of 2016 is 7,45,593 of which 70.22 percent or 5,23,574 are filled up across 31 Departments.

A sanctioned post is a job the government has created and budgeted for, occupied or not. So of 7,45,593 posts on the books, fewer than three in four had a person in them. Against that denominator:

The overall representation of the SC and ST employees of all 31 Departments in comparison with total sanctioned posts comprises of 10.65 per cent and 2.92 per cent respectively.

A share of posts on the books and a share of people actually working are two different measurements, and the report states both. There is a third figure that changes how the first two read:

It noted that 80.35 per cent of the sanctioned posts are concentrated in six major Government departments namely; Education, Home, Health, Revenue, Judicial and Finance.

A statewide average built mostly out of six departments, each running from its lowest posts to its most senior, described by one percentage.

Signed in 2018, switched on in 1995

The law built on that report took an unusual route. Most State Bills need one signature. On 6 December 2017 the Governor decided this one should go to the President, citing the earlier judgment and the importance of the constitutional question,¹ and on 16 December reserved it under Article 200.¹ It was assented to on 14 June 2018 and published in the Gazette on 23 June.¹

Then the same technique the Eighty-fifth Amendment had used at the level of the Constitution, now at the level of a State:

Section 1 (2) provides that the Reservation Act 2018 came into force with effect from 17 June 1995 (the effective date of the seventy-seventh and eighty-fifth constitutional amendments).

Section 4 carried the protection further back still, validating consequential seniority granted under the 1978 order.¹ And Section 9 closed the door behind it:

no suit or other proceedings shall be maintained or continued in any court or any tribunal or before any authority for the review of any such promotions contrary to the provisions of this Act; and (b) no court

An Act signed in 2018, operating from 1995, protecting promotions from 1978, and barring the proceedings through which anyone could ask a court to look at them.

The case against it

The petitioners attacked the evidence and the enactment separately, and these are their submissions as the judgment records them, not findings.

On the evidence, their senior counsel Dr Rajeev Dhavan put it as arithmetic: only thirty-one of sixty-two departments were examined.¹ And the data had been assembled by Groups A, B, C and D rather than by cadre.³ A cadre is the particular service a post sits in and is promoted within; a Group sits above many cadres at once. On that reading the unit of proof and the unit of measurement did not line up.

On the enactment, the charge was sharper:

Dr Dhavan has urged that the Reservation Act 2018 was enacted in a hurry with no purpose other than to overrule the decision in B K Pavitra I, while the issue of implementation was still pending.

And the constitutional name for it: that the State had undertaken an exercise which constituted a clear usurpation of judicial power.¹ Mr Shekhar Naphade added that reaching back to 1978 was arbitrary.¹

2019: cured

The Court did not accept any of it. The challenge to the constitutional validity of the Act was lacking in substance, which is the judgment's own phrase,¹ and the Act was a valid exercise of the enabling power. The finding that carries the judgment is one word:

The Reservation Act 2018 has cured the deficiency which was noticed by B K Pavitra I in respect of the Reservation Act 2002.

The first judgment had said something was missing; this one held the State had supplied it. And the Court went further on the measure itself:

Consequently, it is open to the State to make reservation in promotion for SCs and STs proportionate to their representation in the general population.

That benchmark is the share of these groups in the population at large, not their share inside the service. Two different denominators, capable of two different answers about who is adequately represented.

2022: the counting unit

On 28 January 2022 the Court returned to how the counting had been done, and held that approving collection by groups rather than cadres was contrary to Nagaraj and to Jarnail Singh.³ Count wide and an average can hide a shortfall inside it; count narrow and every service answers for itself.

The standard it set has three parts, and the third is what changes behaviour:

we may make it clear that quantifiable data shall be collected by the State, on the parameters as stipulated in Nagaraj (supra) on the inadequacy of representation, which can be tested by the Courts. We may fur

Data that can be tested by a court is data that has to exist in a form somebody outside the government can open and check. The Court also held that Nagaraj should have prospective effect,³ and it declined to reopen the foundation at all.³

What the Union could show

The last set of figures in this record is not about Karnataka. Asked about cadres across the central government, the Union said there are around 90 Ministries and Departments,³ across which 30 lakh employees work.³ Its estimate of about 3,800 cadres³ rested on what it had been able to gather:

information received from 44 Ministries/Departments out of

Forty-four of ninety. The machinery is not new: instructions for a Deputy Secretary rank Liaison Officer to ensure compliance with reservation instructions have existed since 2013.³

So the position on this record is a standard, a rejected counting unit, and a hearing date: the matters were listed for further hearing on 24 February 2022.³ Nothing in this archive records what happened after it. These proceedings are live, and no outcome is reported here because none is on the record to report.

What is reportable is the shape. In 2006 a State was told to count before it reserved. In 2018 one of the three things it had to count was removed. In 2019 a State that had counted in forty-four days was held to have cured the defect. In 2022 the unit it had counted in was held to be the wrong one, and the Union of India, asked the same question about its own establishment, could answer for forty-four of its roughly ninety Ministries and Departments.

A proof requirement is only as real as the administration's ability to produce the proof. That is the open question, and it is not one a judgment can close.

MangoPOV. "Jarnail Singh and B.K. Pavitra explained: the data test for promotion quotas." 15 September 2026. https://mangopov.in/stories/quantifiable-data-karnataka-ratna-prabha-cadre/

@misc{mangopov-quantifiable-data-karnataka-ratna-prabha-cadre,
  title        = {Jarnail Singh and B.K. Pavitra explained: the data test for promotion quotas},
  author       = {{MangoPOV}},
  year         = {2026},
  howpublished = {\url{https://mangopov.in/stories/quantifiable-data-karnataka-ratna-prabha-cadre/}},
}